Work

Automated Modal Identification via Progressive OPTICS Mode Clustering and Mutually Exclusive True Mode Screening

Year: 2026

Type: article

Abstract: This paper presents an automated method for modal identification for potential applications in long‐term structural health monitoring, combining progressive OPTICS (a clustering algorithm named “ordering points to identify the clustering structure”) mode clustering and mutually exclusive true mode screening strategies. Candidate modes are first extracted using a covariance‐driven stochastic subspace identification algorithm, followed by stabilization diagram cleaning that integrates hard validation criteria (damping ratio and complex conjugate pairs) and soft validation criteria (mean phase deviation and modal energy level). Mode clustering proceeds in two stages: an initial OPTICS clustering based on frequency‐distance constructs stabilization axes, and a final OPTICS clustering based on mode‐shape distance refines the results. To address variations in frequency and mode‐shape distributions, a sliding‐window median threshold strategy and an adaptive bimodal threshold detection strategy are applied sequentially to refine the two OPTICS algorithms. The median damping‐ratio method is then used to extract the representative mode for each cluster. A novel mutually exclusive screening strategy distinguishes true from false clusters, which groups clusters with high mode shape similarity into the same set. Using features such as mode quantity, mean phase deviation, and modal energy level, the k‐means clustering algorithm is applied iteratively within each set to eliminate false clusters, with the representative modes of true clusters serving as the final results. Both numerical and benchmark analyses demonstrate that the method offers several advantages, including the preservation of weak modes, enhanced aggregation of modal parameters, and robustness in key parameters, such as the estimated number of structural modes and the maximum model order. These characteristics reduce modal omission and false modes in practical applications.

Source: Structural Control and Health Monitoring

DOI: 10.1155/stc/1241918

Authors: Xiaoshun Wu, Minsai Zeng, Jun Ma, Cui Li, Muhao Chen

Institutions: Jiangxi University of Science and Technology, Jiangxi Vocational and Technical College of Communications, University of Kentucky

Language: English

Topic: Structural Health Monitoring Techniques

Subfield: Civil and Structural Engineering

Field: Engineering

Domain: Physical Sciences

Open Access status: gold

Estimated APC paid: $2,490

Funder: National Natural Science Foundation of China